SDG News Spotlight: Geoffrey Hinton
Geoffrey Hinton helped build the technology. He now spends his time telling governments that capability is outrunning the institutions meant to control it.
Geoffrey Hinton, University Professor Emeritus at the University of Toronto, shared the 2024 Nobel Prize in Physics with John Hopfield for foundational discoveries and inventions that enabled machine learning with artificial neural networks. His research helped establish the scientific basis for the deep-learning systems now reshaping economies, national security, science and global competition.
Yet at the Nobel banquet in Stockholm on December 10, 2024, Hinton used one of the defining moments of his career to emphasize not the scale of AI’s achievement, but the scale of the responsibility accompanying it.
A Warning From Inside the Revolution
Geoffrey Hinton acknowledged AI’s potential to transform healthcare, education, productivity and scientific discovery. But he also warned of more immediate risks, including cybercrime, surveillance, disinformation, autonomous weapons and potential biological misuse.
His more consequential concern is longer term: that increasingly capable systems could eventually exceed human intelligence without humanity having solved the problem of reliably controlling them.
“We have no idea whether we can stay in control,” Hinton has warned.
That statement captures the central tension now shaping his public work. Hinton is not arguing that a catastrophic outcome is inevitable. He is arguing that the uncertainty itself is dangerous when the systems being developed could become extraordinarily powerful.
This is an important distinction for policymakers. The governance challenge is not dependent on knowing exactly when, or even whether, superhuman AI will emerge. It is whether the consequences are significant enough that governments should prepare before the uncertainty is resolved.
The Race Between Capability and Control
Since leaving Google in 2023, Geoffrey Hinton has increasingly focused on what he sees as an imbalance at the heart of the AI race: investment in making systems more capable is vastly outpacing investment in understanding how to make them reliably safe.
In January 2026, the University of Toronto announced that Hinton would expand his global AI safety work through its Schwartz Reisman Institute for Technology and Society, supported by a $700,000 gift from Good Ventures.
Hinton has framed the problem in stark economic terms, arguing that only a small fraction of corporate AI investment is directed toward safety compared with the resources devoted to increasing model capabilities.
That imbalance points to a broader structural problem. Individual companies face powerful incentives to move faster, particularly as AI becomes strategically important to national competitiveness. Yet many of the most serious risks would extend far beyond any single company: security failures, misuse, systemic instability or loss of control.
In other words, the market may reward acceleration while leaving much of the downside to governments and society.
From Technical Problem to Governance Problem
Hinton’s current role is therefore increasingly less about predicting the exact trajectory of AI and more about forcing governments to confront the institutional architecture surrounding it.
On September 16, Geoffrey Hinton is scheduled to participate in a bipartisan U.S. Senate briefing convened by Senator Bernie Sanders on risks associated with advanced AI systems, alongside researchers including Max Tegmark and Ajeya Cotra.
The significance is not simply that Congress is hearing another warning about AI. It is that the debate is shifting from familiar questions around privacy, bias and misinformation toward more fundamental questions about autonomy, control and systemic risk.
Who should bear the burden of proving increasingly capable systems are safe? Should governments require minimum levels of safety research or independent testing? How should states manage a technology that is simultaneously a commercial asset, a national-security capability and a potentially global source of risk?
And perhaps most difficult: what happens when the incentives of individual companies and countries to move quickly conflict with the collective interest in moving safely?
Why Geoffrey Hinton Matters Now
Hinton’s influence comes from an unusual position. He is neither a regulator looking at AI from the outside nor an industry executive defending a commercial strategy. He is one of the scientists whose work helped make the current technological leap possible.
That gives his warning a particular weight.
But the deeper significance of his message is not that governments should simply fear AI. It is that the existing institutions governing technology were largely designed for risks that emerge after deployment. Advanced AI may require institutions capable of acting before the full scale of the risk is known.
That is a fundamentally different governance problem.
AI development is now being shaped simultaneously by venture capital, technology companies, sovereign competition, defense priorities and scientific ambition. Safety governance, by contrast, remains fragmented across national regulators, voluntary commitments and emerging international initiatives.
Geoffrey Hinton’s warning is ultimately about that asymmetry.
The world is building increasingly powerful intelligence at extraordinary speed. The systems designed to govern that intelligence are evolving much more slowly.
For governments, that may be the most important AI gap of all.
Follow SDG News on LinkedIn
